Engineering Problem-Solving Techniques

Last week, I described four design patterns for AI agentic workflows that I believe will drive significant progress: Reflection, Tool use, Planning and Multi-agent collaboration. Instead of having an LLM generate its final output directly, an agentic workflow prompts the LLM multiple times, giving it opportunities to build step by step to higher-quality output. Here, I'd like to discuss Reflection. It's relatively quick to implement, and I've seen it lead to surprising performance gains. You may have had the experience of prompting ChatGPT/Claude/Gemini, receiving unsatisfactory output, delivering critical feedback to help the LLM improve its response, and then getting a better response. What if you automate the step of delivering critical feedback, so the model automatically criticizes its own output and improves its response? This is the crux of Reflection. Take the task of asking an LLM to write code. We can prompt it to generate the desired code directly to carry out some task X. Then, we can prompt it to reflect on its own output, perhaps as follows: Here’s code intended for task X: [previously generated code] Check the code carefully for correctness, style, and efficiency, and give constructive criticism for how to improve it. Sometimes this causes the LLM to spot problems and come up with constructive suggestions. Next, we can prompt the LLM with context including (i) the previously generated code and (ii) the constructive feedback, and ask it to use the feedback to rewrite the code. This can lead to a better response. Repeating the criticism/rewrite process might yield further improvements. This self-reflection process allows the LLM to spot gaps and improve its output on a variety of tasks including producing code, writing text, and answering questions. And we can go beyond self-reflection by giving the LLM tools that help evaluate its output; for example, running its code through a few unit tests to check whether it generates correct results on test cases or searching the web to double-check text output. Then it can reflect on any errors it found and come up with ideas for improvement. Further, we can implement Reflection using a multi-agent framework. I've found it convenient to create two agents, one prompted to generate good outputs and the other prompted to give constructive criticism of the first agent's output. The resulting discussion between the two agents leads to improved responses. Reflection is a relatively basic type of agentic workflow, but I've been delighted by how much it improved my applications’ results. If you’re interested in learning more about reflection, I recommend: - Self-Refine: Iterative Refinement with Self-Feedback, by Madaan et al. (2023) - Reflexion: Language Agents with Verbal Reinforcement Learning, by Shinn et al. (2023) - CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing, by Gou et al. (2024) [Original text: https://lnkd.in/g4bTuWtU ]

5-WHY ROOT CAUSE ANALYSIS (RCA) Problem Statement: A batch of parts was rejected due to an oversized hole diameter. 5-Why Analysis: 1.Why was the batch rejected?→ Because the hole diameter was larger than the specified tolerance. 2.Why was the hole diameter too large?→ Because the drilling machine was not properly adjusted. 3.Why was the machine not properly adjusted?→ Because the operator used an outdated setup sheet. 4.Why did the operator use an outdated setup sheet?→ Because the latest revision was not available at the machine. 5.Why was the latest revision not available at the machine?→ Because there is no system in place to ensure controlled document distribution. Root Cause: No document control system for distributing updated setup sheets. Corrective Actions: •Introduce a document control procedure to issue and display the latest revision only. •Restrict access to outdated setup sheets by removing old versions from machines. •Train machine operators and line leaders on verifying document revision before setup. Preventive Measures: •Digitize all setup sheets with access through a centralized network folder or MES (Manufacturing Execution System). •Implement revision control logs with sign-off for updates and acknowledgments by operators. •Conduct regular audits on setup documents at workstations. •Establish standard work that includes a revision check step before every job setup. •Integrate barcode or QR code scanning to verify correct document versions at machines.

This is how Anthropic decides what to build next—and it's brilliant. Instead of endless spec documents and roadmap debates, the Claude Code team has cracked the code on feature prioritization: prototype first, decide later. Here's their process (shared by Catherine Wu, Product Lead at Anthropic): Step 1: Idea → Prototype Got a feature idea? Skip the spec. Build a working prototype using Claude Code instead. Step 2: Internal Launch Ship that prototype to all Anthropic engineers immediately. No polish required—just functionality. Step 3: Watch & Listen Track usage religiously. Collect feedback actively. Let real behavior, not opinions, guide decisions. Step 4: Data-Driven Prioritization - High usage + positive feedback → roadmap priority - Low engagement or complaints → back to iteration This "prototype-first product shaping" flips traditional product development on its head. Instead of guessing what users want, they're measuring what users actually use. The beauty? They're dogfooding their own tool to build their own tool. The feedback loop is immediate, honest, and impossible to ignore. The takeaway: Your best product decisions come from real user behavior, not theoretical frameworks. Sometimes the fastest way to validate an idea isn't a survey or interview—it's a working prototype.

Optimizing Thermal Systems: A Deep Dive into Heat Exchanger Calculations for Process Engineers As chemical and process engineers, we often find ourselves at the intersection of theory and industrial application. One of the most critical components in our thermal systems is the heat exchanger — and understanding its calculations is fundamental to efficient, safe, and cost-effective plant design. Here’s a consolidated reference of standard heat exchanger equations that every engineer in the process industry should master: 1. Heat Duty (Q): Q = m × Cp × ΔT Where: m = mass flow rate (kg/s) Cp = specific heat capacity (kJ/kg·K) ΔT = temperature difference between inlet and outlet (K) This equation gives the amount of heat transferred by the fluid and is foundational in energy balance. 2. Log Mean Temperature Difference (LMTD): LMTD = (ΔT₁ - ΔT₂) / ln(ΔT₁ / ΔT₂) Where: ΔT₁ = temp difference at the hot end ΔT₂ = temp difference at the cold end This method is used when both inlet and outlet temperatures are known, ideal for shell & tube or plate exchangers. 3. Overall Heat Transfer Equation: Q = U × A × LMTD Where: U = overall heat transfer coefficient (W/m²·K) A = surface area available for heat exchange (m²) This links the thermal design to the physical parameters of the exchanger. 4. NTU Method (Effectiveness-NTU approach): Used when outlet temperatures are unknown or variable. Effectiveness = Q / Qmax NTU = (U × A) / Cmin Where Cmin is the minimum heat capacity rate among the fluids. These formulas form the core of thermal design, diagnostics, and scale-up. As we aim for energy-efficient, safe, and sustainable operations, mastering these principles becomes non-negotiable. Whether you're involved in equipment design, process simulation, or plant operations, a clear command of heat exchanger fundamentals enables smarter engineering decisions. Let’s continue building better systems, one calculation at a time. #ProcessEngineering #HeatTransfer #ChemicalEngineering #HeatExchangerDesign #EnergyEfficiency #EngineeringExcellence #ThermalSystems #PlantDesign #ProcessOptimization

I could watch this 19th-century water-powered sash sawmill run all day. No electronics. No PLCs. No hydraulics. Just gravity, flowing water, wood, iron, and some very clever mechanical engineering. What impresses me most is not just that it works—it’s that it still works. The cams, linkages, and wooden frame are all doing exactly what they were designed to do over a century ago: convert the steady force of moving water into a precise, repeatable cutting motion. Designed without CAD, built without CNC, and yet the tolerances, alignments, and load paths are good enough to survive generations of real-world use. It is a masterclass in: -Simplicity of design -Durability and maintainability -Using local materials and available energy -Engineering that respects both physics and craftsmanship In modern projects we talk a lot about “sustainability,” “resilience,” and “design for maintenance.” This old mill is a reminder that those ideas are not new. The millwrights, carpenters, and blacksmiths who built systems like this were solving the same problems we face today—just with different tools. As engineers, inspectors, and builders, there is a lot we can still learn from these legacy systems about robustness, clarity of design, and respect for the trades that bring our drawings to life. #engineering #manufacturing #mechanicalengineering #craftsmanship #industrialhistory #design #quality

How do engineers simulate turbulence? Let’s explain it like you’re 5. Imagine you’re watching a big football match. The players are running everywhere, passing the ball, the crowd is cheering. It's total chaos sometimes. Now, you have three ways to understand what’s happening on the field: 1. DNS (Direct Numerical Simulation): You follow every single player, every step, every pass, every move of the ball, every bounce. You know exactly what’s happening. But it's exhausting and takes a super powerful camera and endless storage! ➡️ In fluid flow, DNS resolves every little swirl and eddy (smallest to largest). Super accurate but takes huge computing power and storage. 2. LES (Large Eddy Simulation): You follow only the star and big plays, like goals and key passes. For the small moves and background players, you just make a smart guess. You still understand most of the game, and it's easier to watch. You might miss some tiny details. ➡️ In fluid flow, LES resolves the big turbulent eddies and models the small ones. A balance between detail and effort. 3. RANS (Reynolds-Averaged Navier-Stokes): You don’t watch every move. You just say, “On average, this team had more possession and scored more goals.” Fast and easy to understand the overall result. But you miss all the exciting plays and details. ➡️ In flow, RANS gives the average effect of turbulence on the mean flow, without tracking all the chaos. It all depends on what you need. The full match analysis? (DNS) The highlights? (LES) Or just the final score? (RANS) That’s how engineers balance accuracy and computing power in turbulence modeling. #mechanical #aerospace #turbulence #automotive #cfd #aerodynamics

Chemical Engineering interview question on Heat Exchanger Design Question : Can you walk me through the complete steps for designing a heat exchanger? Answer : 1. Define the Objective and Application What is the heat exchanger supposed to do? Heating, cooling, condensing, or vaporizing? What type is suitable—shell and tube, plate, air-cooled, spiral? Collect Process Data -Fluid types (shell side and tube side) -Inlet and outlet temperatures -Flow rates (mass or volumetric) -Pressure limits, phase (liquid/gas), corrosiveness -Fouling factors. 2. Perform Energy Balance Heat duty (Q) is calculated as: Q = m × Cp × ΔT (for single-phase fluids) Q = m × λ (for phase change, e.g., condensation or evaporation) 3. Select Flow Configuration Counterflow, parallel flow, crossflow Affects log mean temperature difference (LMTD) Determine Log Mean Temperature Difference (LMTD) For counterflow or parallel flow: ΔTlm = (ΔT1 - ΔT2) / ln(ΔT1 / ΔT2) where ΔT1 and ΔT2 are temperature differences at each end Apply correction factor (F) for multi-pass or crossflow: Q = U × A × ΔTlm × F 4. Estimate Overall Heat Transfer Coefficient (U) Use the thermal resistance model: 1/U = 1/hi + Rf1 + x/k + Rf2 + 1/ho where: hi and ho = inside and outside heat transfer coefficients Rf1 and Rf2 = fouling resistances x = wall thickness k = thermal conductivity of wall material 5. Calculate Required Heat Transfer Area (A) A = Q / (U × ΔTlm × F) Design Tube and Shell Geometry Decide: Tube length, outer diameter, pitch, layout (triangular or square) Number of tube passes Shell diameter and baffle spacing 6.Pressure Drop Calculations - For tube side: ΔP = f × (L/D) × (ρv² / 2) -For shell side: use Bell-Delaware method or Kern method based on layout 7. Check Velocity Limits Ensure velocity is within range to avoid erosion and ensure turbulence Typically: 1 to 2.5 m/s in tubes, 0.3 to 1 m/s in shell Material Selection Based on corrosion resistance, temperature, pressure, and cost Common materials: SS316, carbon steel, copper alloys, titanium 8. Mechanical Design and Code Compliance Comply with ASME Section VIII, TEMA standards Check for allowable stress, corrosion allowance, gasket types, expansion allowances Cleaning and Maintenance Consideration Tube side often chosen for fouling fluids due to easier access Decide on removable bundle, floating head, or fixed tube sheet design 9. Cost Optimization Optimize based on surface area, pressure drop, and maintenance frequency Consider life cycle cost, not just capex 10. Final Review Simulate in process design tools (e.g., HTRI, Aspen EDR) Cross-check thermal and mechanical integrity Validate against client and safety requirements

Stop loading data into ChatGPT and asking for insights. It is lying to you. An LLM cannot find "truth." It does not know your business context. It does not understand your data. It fabricates plausible narratives and reinforces your confirmation bias. You don't need an insight generator. You need a sparring partner. The LLM's true power is in stress-testing your ideas. This is how you shatter bias. This is how you find the real insight, not just the one you were looking for. Use the "Challenge-Code-Verify" cycle. - The Challenge: State your hypothesis. Command the LLM to act as a skeptical statistician and find 3 ways you are wrong. - The Code: Direct the LLM to produce the exact code (Python/R) or formula (Excel/Sheets) to test its counter-argument. - The Verification: Run the code. Look at the chart. Make the call. This is how you partner with the LLM to sharpen your human abilities -- intuition, creativity, novelty. Asking your LLM for insights is like asking your sparring partner to fight for you. It will get knocked out. Its job isn't to win the match. Its job is to reveal your weaknesses, sharpen your skills, perfect your form, and force you to be better. Spar with your LLM so that when its showtime, you are the one who lands the knockout. Art+Science Analytics Institute | University of Notre Dame | University of Notre Dame - Mendoza College of Business | University of Illinois Urbana-Champaign | University of Chicago | D'Amore-McKim School of Business at Northeastern University | ELVTR | Grow with Google - Data Analytics #Analytics #DataStorytelling

Can quantum computing revolutionize computational mechanics? In our paper "Towards Quantum Computational Mechanics", we introduce a PDE solver that achieves exponential speedup, reducing the complexity of representative volume element (RVE) computations from O(Nᶜ) in classical computing to O((log N)ᶜ). This exponential acceleration over classical solvers brings concurrent multiscale computing one step closer to practicality. https://lnkd.in/ebxTBG4Z Our research, recently accepted in Computer Methods in Applied Mechanics and Engineering, is a joint effort by Burigede Liu, Michael Ortiz, and myself.